The Göteborg 70 Stroke Study

Transcription

1 Resource Utilization and Costs of Stroke Unit Care Integrated in a Care Continuum: A 1-Year Controlled, Prospective, Randomized Study in Elderly Patients The Göteborg 70 Stroke Study Lisbeth Claesson, OTR, MSc; Gunilla Gosman-Hedström, OTR, MSc; Magnus Johannesson, PhD; Björn Fagerberg, MD, PhD; Christian Blomstrand, MD, PhD Background and Purpose The aim of the present study was to examine resource utilization during a 12-month period after acute stroke in elderly patients randomized to care in an acute stroke unit integrated with a care continuum compared with conventional care in general medical wards. A secondary aim was to describe costs related to the severity of stroke. Methods Two hundred forty-nine consecutive patients aged 70 years with acute stroke within 7 days before admission, living in their own homes in Göteborg, Sweden, without recognized need of care were randomized to 2 groups: 166 patients were assigned to nonintensive stroke unit care with a care continuum, and 83 patients were assigned to conventional care. There was no difference in mortality or the proportion of patients living at home after 1 year. Main outcomes were costs from inpatient care, outpatient care, and informal care. Results Mean annual cost per patient was Swedish crowns (SEK) (equivalent to $25 373) and SEK ($28 507) in the stroke unit and the general medical ward groups, respectively (P NS). Seventy percent of the total cost was for inpatient care, and 30% was for outpatient and informal care. For patients with mild, moderate, and severe stroke, the mean annual costs per patient were SEK ($15 970), SEK ($39 254), and SEK ($32 836), respectively (P 0.001). There was no statistical difference in age or nonstroke diagnosis. Conclusions The total costs the first year did not differ significantly between the treatment groups in this prospective study. The total annual cost per patient showed a very large variation, which was related to stroke severity at onset and not to age or nonstroke diagnoses. Costs other than those for hospital care constituted a substantial fraction of total costs and must be taken into account when organizing the management of stroke patients. The high variability in costs necessitates a larger study to assess long-term cost effectiveness. (Stroke. 2000;31: ) Key Words: costs and cost analysis elderly stroke units Sweden Stroke is a major public health problem associated with high mortality, disability, and cost. Therefore, it is important to develop effective treatment strategies for this health problem. The Stroke Unit Trialists Collaboration 1 concludes from the meta-analyses data that stroke unit care improves survival and health status after a stroke. 1 3 However, there is a lack of evidence concerning economic evaluations of stroke unit care versus conventional care. 1 It has been emphasized that economic evaluations of stroke care are not easy to perform because of the complexity of costs. 4 This increases the need for prospective studies with reliable data concerning the impact of stroke unit care. Thus, when cost implications of acute stroke unit care are examined, many factors in the study design may improve the quality of the outcome measures, eg, screening procedure of background population when recruiting patients, randomization, prospective approach to data collection, and inclusion of all costs related to the studied patients. So far, no study has considered all these aspects, and only one research group 5,6 has studied the costs of stroke unit care; however, that study did not use a randomization procedure. The other studies have not been focused on stroke unit care Some studies have been based on data collected retrospectively and not prospectively. 14,19,21 Many reports do not provide information on costs for outpatient care, 5,6,8,10,13,14,19,25 social services, 5,6,8,10,13,14,19,25 and informal care. 5 21,23 26 In some studies, data on costs have not been registered for individual patients but for categories of patients. 8,16,18,23 Received June 24, 2000; final revision received June 24, 2000; accepted June 29, From the Institute of Clinical Neuroscience, Neurological Disease Section (L.C., G.G.-H., C.B.), and the Department of Medicine (B.F.), Sahlgrenska University Hospital, and the Department of Occupational Therapy and Physiotherapy, College of Health and Caring Science (L.C., G.G.-H.), Göteborg University, Göteborg, Sweden, and the Centre for Health Economics, Stockholm School of Economics (M.J.), Stockholm, Sweden. Correspondence to Prof Christian Blomstrand, Institute of Clinical Neuroscience, Neurological Disease Section, Sahlgrenska University Hospital, SE Göteborg, Sweden. cbl@neuro.gu.se 2000 American Heart Association, Inc. Stroke is available at

2 2570 Stroke November 2000 We have performed a 1-year randomized study comparing treatment outcome at an acute stroke unit with a care continuum with the outcome of conventional treatment and have prospectively collected all types of costs for these patients. A care continuum implies an acute stroke unit linked to continued care in geriatric stroke units when a long rehabilitation period is needed. The clinical results have been reported elsewhere and showed no overall significant difference in proportions of patients living at home or in patient mortality or functional ability between the 2 treatment groups after 1 year. 27 However, our data indicate that stroke unit care reduced mortality or the need for institutional care in the short-term perspective, especially among those with severe stroke (95% CI 32% to 9%) or in patients with cardiac disease (95% CI 40% to 3%), supporting the results of the Stroke Unit Trailists meta-analysis 1 and a previous Swedish study. 28 If the health effects are identical for the 2 treatments, an economic evaluation can be carried out as a comparison of the total costs of the treatments. Such an analysis is often referred to as a cost-minimization analysis in the health economics literature. 29 The primary aim of the present study was to identify and analyze costs associated with the treatment of elderly patients with acute stroke by using the experimental setup of a randomized study of care in a stroke unit and conventional care referred to above. Detailed data on resource consumption were collected prospectively during the study. A secondary aim was to estimate the total costs of stroke during the first year after its occurrence in relation to the severity of the stroke. Subjects and Methods The design was a randomized prospective study with 2 parallel groups followed for 1 year in Sahlgrenska University Hospital in Göteborg, Sweden, the Göteborg 70 Stroke study. Patients aged 70 years (n 249) were randomized after informed consent in connection with acute admission to a stroke unit (SU group) integrated with a care continuum to geriatric wards or to a general medical ward (GW group) between February 1, 1993, and May 17,1994. The randomization was a 2:1 ratio of stroke unit to conventional treatment at general medical wards. 30 A 2:1 randomization to the SU groups was applied to obtain a continuous input of patients to these units. At the acute hospital, the stroke unit consisted of 2 wards, 1 in a general medical ward and 1 in a neurological ward. The SU group was composed of 166 patients; the GW group was composed of 83 patients. There were no significant differences between the groups at entry with regard to sex: 66% (n 110) of the SU group and 54% (n 45) of the GW group were women. The mean age of the SU and GW groups was 80.1 and 79.7 years, respectively. There was no difference in living conditions: 59% (n 96) of the SU group and 48% (n 39) of the GW group were living alone. Concerning the medical history, there were no differences; however, angina pectoris was significantly (P 0.05) more common in the SU group. Furthermore, the groups were comparable with regard to neurological score at entry, with a median score of 45 in the SU group and 46 in the GW group. The side of predominant neurological deficit was the right side for 45% (n 74) in the SU group and 42% (n 35) in the GW group and the left side for 51% (n 85) and 45% (n 37), respectively. Speech disorder occurred in 48% (n 79) of the SU group and 49% (n 41) of the GW group. Two patients in each group did not fulfill the criteria for acute cerebrovascular disease. 27 Intervention A nonintensive stroke unit 28,31 was organized in a care continuum with 2 acute stroke units and 2 stroke units at geriatric wards working according to identical principles. The treatment program was built on the principle of stroke unit care with a teamwork concept for nursing and rehabilitation. The program was standardized with regard to diagnostic evaluation, observation, acute treatment, mobilization, and rehabilitation. All staff at the acute wards and at the geriatric wards had a continuous educational program. The acute and geriatric stroke units collaborated in terms of treatment principles, training, and work procedures. Support and information to relatives was an important part of the program, as was the focus on the patients needs and participation in the treatment. Patients in need of prolonged rehabilitation were transferred to geriatric stroke units. The patients in the general ward received conventional acute medical care, physiotherapy, and occupational therapy, although not within the framework of a structured stroke unit care approach. At the stroke unit, the patients received significantly more occupational and physical therapy. 27 Measurements Two registered occupational therapists not involved in the design of the study and the treatment of the patients made all evaluations of the study outcome. The assessments were performed 0 to 3 days after randomization and after 3 weeks, 3 months, and 12 months. The assessments were made at the hospital, in nursing homes, and/or in the patients homes by means of interviews. If there were any doubts about ability, the patients were asked to perform the activity. Patients were randomly assigned to the occupational therapists, who evaluated them during follow-up. The occupational therapists were experienced and welltrained in assessing activities of daily living (ADL). 27 Established methods were used to assess neurological status (Scandinavian Stroke Study Group 32 ), the ability to perform daily life activities (Barthel index 33 and Sunnaas index of ADL 34,35 ), and health-related quality of life (Nottingham Health Profile ). There were no differences between the groups regarding survival. The proportion of patients at home or in institutions did not differ between the groups after 3 or 12 months. At the 12-month assessment, 73% of the SU group and 77% of the GW group were alive, and 63% and 60%, respectively, were living at home. There were no significant differences between the groups regarding neurological score, ADL scores, or the 6 dimensions of the Nottingham Health Profile questionnaire at the 12-month assessment. 27 The study was approved by the ethics committee of the Sahlgrenska University Hospital. Costs All costs were estimated from the time of randomization to the end of the 12-month follow-up. The costs were divided into costs for hospitalization, institutionalized living, outpatient care, different kinds of support, and informal care provided by relatives. The costs of each of these units are shown in Table 1. The total cost was obtained by taking the sum of all separate units and multiplying them by their costs. All costs were estimated according to 1996 prices in Swedish crowns (SEK; exchange rate in 1996, $ SEK and SEK). Data on all hospitalizations during the study were collected from hospital records. The costs were estimated separately for the initial hospitalization, and a distinction was made between acute hospitalization and nonacute hospitalization. For nonacute hospitalization (geriatric wards and postcare unit), unit costs per hospital day were taken from estimations made by the civic administration of the city of Göteborg. For acute hospitalization, the cost was divided into a hotel cost per hospital day and patient-related costs. The hotel cost per hospital day at different wards was based on the hotel costs at the Sahlgrenska University Hospital and included staff costs, rent costs, and overhead costs for food, drugs, cleaning, washing, and transportation. Patient-related costs included medical examinations and treatments related to each individual patient. The internal transfer payments used at the Sahlgrenska University Hospital were considered (as judged by the hospital administration to reflect the actual

3 Claesson et al Cost of Stroke Unit vs Conventional Care 2571 TABLE 1. Unit Costs Used in Analysis, 1996 Prices (SEK) Cost per, SEK Initial hospitalization for index stroke* Stroke unit, hotel cost per day 2987 General ward, hotel cost per day 2671 Geriatric ward, per day 1900 Acute hospitalizations after initial hospitalization to 12 mo Medicine ward, hotel cost per day 2473 Cardiology ward, hotel cost per day 3279 Orthopedic ward, hotel cost per day 2432 Surgery ward, hotel cost per day 3324 Neurology ward, hotel cost per day 2659 Audiology ward, hotel cost per day 2985 Oncology ward, hotel cost per day 2432 Gynecology ward, hotel cost per day 2245 Nonacute hospitalizations from Geriatric ward, per day 1900 Postcare unit, per day 1021 Institutionalized living from Nursing home, per day 898 Home for the elderly, per day 668 Assisted living, per day 441 Outpatient care from Physician visit, per visit 500 Other medical visit, per visit 320 Outpatient rehabilitation, per day 444 Anticoagulation treatment, per visit 142 Different kinds of support from Home assistance, per hour 187 Taxi service for the disabled, per 1-way trip 69 Safety alarm, per month 220 Other costs from Informal care, per hour 38 The costs of prescription drugs, assistive devices, and housing adaptations were also estimated, but the unit costs for these resources are not shown (because the prices varied between different drugs, different types of assistive devices, and different types of housing adaptations; see the text for the estimation of these costs). *The cost for initial hospitalization and costs of acute hospitalizations from were divided into hotel costs and patient-related costs. Unitary costs used to estimate patient-related costs are not shown because of the large number of unit costs used in these estimations. See text for estimation of patient-related costs. Hotel cost per day at the stroke unit includes hotel cost per day at a general ward plus costs for an occupational therapist, a physiotherapist, and a nurse on the stroke team. Nurse, occupational therapist, physiotherapist, speech therapist. costs) as unit costs for these medical examinations and treatments. The stroke unit in the internal medicine ward had a stroke nurse (full time), an occupational therapist (part time), and a physiotherapist (part time) as extra medical staff in the team. The cost of this extra medical staff was added to the hotel cost at this stroke unit. The number of days of institutionalized living (nursing home, home for the elderly, and assisted living) was obtained from hospital records. Unit costs per day were taken from estimations made by the civic administration of the city of Göteborg. Structured interview questionnaires at 3- and 12-month assessments were used to collect data concerning use of outpatient care. If the client was unable to answer, the questions were posted to relatives and/or medical staff. Outpatient care included the following: use of prescription drugs; visits to physicians, nurses, occupational therapists, physiotherapists, and/or speech therapists; outpatient rehabilitation; and visits for anticoagulation treatment. Unit costs for these types of visits were taken from estimations made by the civic administration of the city of Göteborg. To estimate the cost of prescription drugs, official Swedish retail prices were used. 39 In the structured interview questionnaires at 3 and 12 months, data about different kinds of support were also collected. Data were collected concerning hours of home assistance, the number of 1-way taxi trips for disabled people, and the use of safety alarms, assistive devices, and housing adaptations required after stroke. From estimations made by the civic administration of the city of Göteborg, the unit costs for these resources were derived. Costs for housing adaptations and assistive devices varied according to the type of housing adaptation and assistive devices. The final cost component estimated was the informal care provided by relatives. The number of hours of informal care was collected in the structured interview questionnaires at 3 and 12 months. Thirty-eight SEK per hour was used as the unit cost of informal care. This corresponds to 35% of the gross wage rate, and this cost has been used as the cost of leisure time in previous economic evaluations of health care in Sweden. 40,41 No costs for outpatient care, different forms of support, and informal care were estimated for patients who died before 3 months. For patients who died 3 to 12 months after randomization, the units from the 3-month assessment were used to estimate the consumption of these utility resources until the date of death. Statistical Analyses The analyses were performed according to the intention-to-treat principle. 42 Differences in use of resources and costs were compared by use of a Mann-Whitney nonparametric test for continuous variables, and a x 2 test was used for categorical variables. In addition to testing for whether the costs differed between the 2 treatment groups, a Kruskal-Wallis test was also made to determine whether the costs differed between mild, moderate, and severe stroke. 43 The classification of stroke into mild, moderate, and severe was based on the Barthel index score within the first 3 days after randomization according to a definition used by the Stroke Unit Trialists Collaboration. Mild, moderate, and severe stroke were defined as Barthel scores 50 to 100, 15 to 45, and 0 to 10, respectively. 44,45 There were no differences in the number of patients with mild, moderate, and severe stroke allocated to the stroke unit and general ward. A value of P 0.05 (2-sided) was considered significant. Results The use of resources per patient during the study are shown in Table 2, and the costs per patient during the study are shown in Table 3. The costs did not differ significantly between the groups for any of the cost components. The mean total costs per patient were numerically somewhat lower for the SU patients than for the GW patients ( SEK [$25 373] versus SEK [$28 657], respectively), but the difference was not significant. The total costs per patient ranged from 5032 SEK ($751) to SEK ($ ) for the SU patients and from 5661 SEK ($845) to SEK ($ ) for the GW patients. Table 4 shows the costs for mild, moderate, and severe stroke. The mean total costs per patient were SEK ($15 970) for mild stroke, SEK ($39 254) for moderate stroke, and SEK ($32 836) for severe stroke. There was a significant difference (P 0.001) in total costs

4 2572 Stroke November 2000 TABLE 2. Use of Resources During Study SU Group (n 166) GW Group (n 83) Mean Median Percentile Mean Median Percentile P * Initial hospitalization for index stroke Acute hospital, days per patient Geriatric ward, days per patient Acute hospitalizations from, days per patient Nonacute hospitalizations from Geriatric ward, days per patient Postcare unit, days per patient Institutionalized living from Nursing home, days per patient Home for the elderly, days per patient Assisted living, days per patient Outpatient care from Physician visits per patient Other medical visits per patient Outpatient rehabilitation, days per patients Anticoagulation treatment, visit per patient Different kinds of support from Home assistant, hours per patient Safety alarm, months per patient Taxi service for disabled, 1-way trip per patient Assistive devices, no. of devices per patient Housing adaptations, % of patients with modifications Other costs from Informal care, hours per patient Data on consumption of prescription drugs were also collected in study, but quantity of each drug consumed is not shown because of the variety of drugs used (see text for estimation of cost of prescription drugs). *Mann-Whitney test was used for all variables, except for housing adaptations, for which the 2 test was used. Data on use of patient-related resources were also collected at initial hospitalization and at acute hospitalizations from nths. Because of the many different resources used, these resources are not shown in table. See text for estimation of patient-related costs. Nurse, occupational therapist, physiotherapist, speech therapist. for mild, moderate, and severe stroke. The total costs for mild stroke were lower than the costs for moderate and severe stroke, and the highest total costs were shown in moderate stroke. The bulk of the lower costs of mild stroke were the result of lower costs at the initial hospitalization and lower costs of institutionalized living. There were no statistical differences between patients with mild, moderate, or severe stroke in age, sex, living arrangements, or nonstroke diagnosis (Table 5). A problem in the comparison of the costs of mild, moderate, and severe stroke during the first year after the stroke was that the mortality differed for these groups. After 1 year, 13%, 34%, and 45% of the patients had died in the mild, moderate, and severe groups, respectively. Because mortality differed depending on the severity of the stroke, we also estimated the costs for the patients that were alive at the end of the study; these results are shown in Table 6. For patients who survived, the mean total costs per patient were SEK ($15 075) for mild stroke, SEK ($42 239) for moderate stroke, and SEK ($49 403) for severe stroke. There was a significant difference (P 0.001) in total costs for mild, moderate, and severe stroke. The total costs for mild stroke were lower than the costs for moderate and severe stroke, and the highest total costs were shown in severe stroke. Discussion To our knowledge, this is the first randomized controlled trial of acute stroke unit care with a care continuum that prospectively collected comprehensive data about resource consumption in elderly patients with acute stroke. There was no effect on the number of patients living at home after 1 year, but a beneficial effect was found after 3 months of stroke unit care on mortality and need of institutional care among those with concomitant heart disease. There were no significant differences between the groups regarding neurological score, ADL

5 Claesson et al Cost of Stroke Unit vs Conventional Care 2573 TABLE 3. Cost per Patient During Study Cost per Patient, SEK SU Group (n 166) GW Group (n 83) Mean Median Percentile Mean Median Percentile P* Initial hospitalization for index stroke Acute hospital, hotel cost Acute hospital, patient-related costs Geriatric ward Total Acute hospitalizations from Hotel cost Patient-related costs Total Nonacute hospitalizations from Geriatric ward Postcare unit Total Institutionalized living from Nursing home Home for the elderly Assisted living Total Outpatient care from Physician visit Other medical visits Outpatient rehabilitation days Anticoagulation treatment visits Prescription drugs Total Different kinds of support from Home assistant Safety alarm Taxi service for disabled Assistive devices Housing adaptations Total Other costs from Informal care Total costs SEK exchange rate in 1996 was $ SEK and SEK. *Mann-Whitney test was used to test significance of all variables. Nurse, occupational therapist, physiotherapist, speech therapist. scores, or the 6 dimensions of the Nottingham Health Profile questionnaire at the 12-month assessment. 27 During the 12 months of follow-up, detailed data about inpatient care, institutional living, outpatient care, social services, and informal care were collected. According to the results of the present study, the total costs per patient did not differ significantly between patients randomized to a stroke unit and patients randomized to a general medicine ward. A secondary aim of the present study was to estimate the costs of stroke in relation to the severity of the stroke (mild, moderate, and severe stroke). This analysis showed that the costs increase with the severity of the stroke; this was not explained by differences in age or in comorbidity. This is in accordance with Jörgensen et al, 6 who found that comorbidity did not increase the length of hospital stay in stroke patients. The costs were substantially lower among patients with mild stroke. The lower costs at the initial hospitalization and the lower costs of institutionalized living explain the total lower costs of mild stroke. A previous study has demonstrated a similar finding, although the comorbidity was not reported. 7

6 2574 Stroke November 2000 TABLE 4. Cost per Patient During Study According to Mild, Moderate, and Severe Stroke Cost per Patient, SEK Mild Stroke (n 111) Moderate Stroke (n 59) Severe Stroke (n 75) Mean Median Percentile Mean Median Percentile Mean Median Percentile P* Initial hospitalization for index stroke Acute hospital, hotel cost Acute hospital, patient-related costs Geriatric ward Total Acute hospitalizations from Hotel cost Patient-related costs Total Nonacute hospitalizations from Geriatric ward Postcare unit Total Institutionalized living from Nursing home Home for the elderly Assisted living Total Outpatient care from Physician visit Other medical visits Outpatient rehabilitation Anticoagulation treatment visits Prescription drugs Total Different kinds of support from Home assistance Safety alarm Taxi service for disabled Assistive devices Housing adaptations Total Other costs from discharge to 12 mo Informal care Total costs SEK exchange rate in 1996 was $ SEK and SEK. *Kruskal-Wallis test was used to test significance of all variables. Nurse, occupational therapist, physiotherapist, speech therapist. The costs per patient varied greatly depending on the severity of stroke; they also varied within each severity class. This great variability in costs affected the statistical power to find a significant difference in costs between the treatment groups. The power calculations of the present study, and thus the sample size, were based on detecting a difference in the proportion of patients discharged to their own homes. The result of the present study can be used for sample-size calculations of future studies of the costs of stroke. Our data indicate that 400 patients must be recruited to obtain sufficient statistical power to show a 25% reduction in costs the first year after stroke. The present study included 249 patients, and it had a limited statistical power in demonstrating a significant difference in costs between the treatment groups. An advantage of this randomized study was that patientspecific data on both costs and outcomes were collected

7 Claesson et al Cost of Stroke Unit vs Conventional Care 2575 TABLE 5. Baseline Characteristics According to Mild, Moderate, and Severe Stroke Mild Stroke (n 111) Moderate Stroke (n 59) Severe Stroke (n 75) Female sex, n (%) 69 (62) 39 (66) 45 (60) Age (mean), y Living alone, n (%) 58 (53) 37 (65) 38 (51) Medical history,* n (%) Myocardial infarction 22 (20) 10 (17) 14 (19) Intermittent claudication 2 (2) 4 (7) 6 (8) Atrial fibrillation 26 (23) 13 (22) 23 (31) Chronic heart failure 19 (17) 6 (10) 11 (15) Diabetes mellitus 28 (25) 10 (17) 28 (25) Hypertension 42 (38) 19 (32) 26 (35) Angina pectoris 29 (26) 14 (24) 13 (17) Cancer 16 (14) 9 (15) 9 (12) Chronic respiratory disease 4 (4) 6 (10) 6 (8) Gastrointestinal disease 7 (6) 1 (2) 4 (5) Other diagnosis 12 (11) 10 (17) 17 (23) No other diagnosis 67 (60) 34 (58) 44 (59) *Patients can be represented in several of the diagnosis groups. prospectively, indicating a high internal validity. Increased reliability was considered by the fact that the 2 occupational therapists responsible for evaluations were experienced and independent with respect to the treatment of the patients. All data on resource consumption were also scrupulously collected prospectively within the trial. It has been pointed out that a weakness in this type of study could be a lower degree of external validity. 29 However, this randomized study has been based on generally accepted clinical practice, which decreases the problem of external validity. The patients recruited to the study are also likely to be representative of elderly patients treated for stroke in Sweden. An alternate method might be to predict long-term costs from functional status. 46 Lifetime costs for persons suffering from stroke is another method of expressing costs. 16,18 There are also studies that have estimated costs for stroke care during 1 calendar year. 8,23 The present data are not directly comparable to the results of previous studies, because the present study is the only one to have used a design of prospectively collected data on all types of costs in a randomized clinical trial based on a representative elderly stroke population. Accordingly, these results cannot be directly compared with any previous studies. However, many previous studies have estimated the costs of stroke Most of them have concentrated on the costs of inpatient care and institutional living, 5,6,8,10,13,14,19 21,25 and costs for outpatient care, social services, and informal care were not included. The total costs found in the present study are of the same order as those estimated in a thorough analysis of costs of stroke in Sweden, 11 but the relative costs for nonhospital care are higher in the present study. The present study showed that inpatient care and institutional living constituted 70% of the total costs for patients randomized to the SU group and 71% of the total costs for patients randomized to the GW group. Twenty percent of the total costs in the SU group were costs of outpatient care, and 10% were costs of informal care. In the GW group, these shares were 16% and 13%, respectively. Thus, one third of the total costs during the year of follow-up were costs that were not associated with inpatient care and institutionalized living, which indicates the importance of comprehensive prospective studies on stroke care. It is also likely that the fraction of these costs would increase further with longer followup, because the initial hospitalization determined a large fraction of the costs during the first year after the stroke. It should also be pointed out that the costs of informal care were conservatively estimated, inasmuch as it was assumed that the informal care reduced the leisure time rather than the working time of the caregivers. Leisure time was also valued rather conservatively at 38 SEK per hour (35% of the gross wage rate). Doubling the value of leisure time to 76 SEK per hour would increase the share of total costs for informal care to 20%. The present study showed that the total costs after stroke increased with the severity of the stroke. A possibly better stroke unit effect for patients with severe stroke corresponds with the findings of other studies, which have shown that the severity of stroke influences the length of hospital stay 6,21 and the pattern of total utilization of both hospital and nonhospital care. 7,17 Furthermore, Stroke Unit Trialists Collaborations meta-analytic data indicate a more marked effect of stroke unit care for patients with severe stroke versus mild stroke. 44 In conclusion, the present study showed that the total costs the first year did not differ significantly between the treatment groups. The total annual cost per patient showed a very large variation, which was related to stroke severity at onset and not to age or nonstroke diagnoses. Costs other than those for hospital care constituted a substantial fraction of total costs and must be taken into account when organizing the management of stroke patients. The high variability in costs necessitates a larger study to assess long-term cost effectiveness. Acknowledgments This study was supported by the Vårdal Foundation, Trygghetsfonden, the Swedish Stroke Association, John and Brit Wennerström s Foundation for Neurological Research, Felix Neuberg Foundation, Rune and Ulla Amlöv s Foundation for Neurological Research, Hjalmar Svensson Research Foundation, and King Gustav V and Queen Viktoria Foundation. Prof Birgitta Lundgren-Lindquist (present address La Trobe University, Melbourne, Australia) generously provided us with resources and support. We are grateful for the cooperation of Dr Ulla Hallhagen (Högsbo Hospital) and Dr Michaela Holmdahl (previously Vasa Hospital, now Department of Geriatrics, Sahlgrenska University Hospital). We would like to thank Dr Lena Bokemark (Department of Medicine), who helped us to collect necessary data from the medical records. We would also like to thank Agneta Hallén and Henrik Siverbo, who helped us with the data administration, and Stefan Granbom for valuable help with statistics. A special thanks to the staff at the stroke units and the physicians and staff at the Emergency Department who assisted in randomizing the patients.

8 2576 Stroke November 2000 TABLE 6. Cost per Patient Who Survived During Study According to Mild, Moderate, and Severe Stroke Cost per Patient, SEK Mild Stroke (n 97) Moderate Stroke (n 45) Severe Stroke (n 41) Mean Median Percentile Mean Median Percentile Mean Median Percentile P* Initial hospitalization for index stroke Acute hospital, hotel cost Acute hospital, patient-related costs Geriatric ward Total Acute hospitalizations from Hotel cost Patient-related costs Total Nonacute hospitalizations from Geriatric ward Postcare unit Total Institutionalized living from Nursing home Home for the elderly Assisted living Total Outpatient care from discharge to 12 mo Physician visit Other medical visits Outpatient rehabilitation Anticoagulation treatment visits Prescription drugs Total Different kinds of support from Home assistance Safety alarm Taxi service for disabled Assistive devices Housing adaptations Total Other costs from discharge to 12 mo Informal care Total costs SEK exchange rate in 1996 was $ and SEK. *Kruskal-Wallis test was used to test significance of all variables. Nurse, occupational therapist, physiotherapist, speech therapist.

9 Claesson et al Cost of Stroke Unit vs Conventional Care 2577 References 1. Langhorne P, Dennis M. Stroke Units: An Evidence-Based Approach. London, UK: BMJ Books; Stroke Unit Trialists Collaboration. How do stroke units improve patients outcomes?: a collaborative systematic review of randomized trials. Stroke. 1997;28: Stroke Unit Trialists Collaboration. Collaborative systematic review of the randomised trials of organised in-patient (stroke unit) care after stroke. BMJ. 1997;314: Gladman JRF. Stroke units: are they cost effective? Br J Hosp Med. 1992;47: Jörgensen HS, Nakayama H, Raaschou HO, Larsen K, Hubbe P, Skyhöj Olsen T. The effect of stroke unit: reductions in mortality, discharge rate to nursing home, length of hospital stay, and cost: a community-based study. Stroke. 1995;26: Jörgensen HS, Nakayama H, Raaschou HO, Skyhöj Olsen T. Acute stroke care and rehabilitation: an analysis of direct cost and its clinical and social determinants: the Copenhagen Stroke Study. Stroke. 1997;28: Thorngren M, Westling B. Utilization of health care resources after stroke: a population-based study of 258 hospitalized cases followed during the first year. Acta Neurol Scand. 1991;84: Isard PA, Forbes JF. The cost of stroke to the national health service in Scotland. Cerebrovasc Dis. 1992;2: Young J, Forster A. Day hospital and home physiotherapy for stroke patients. a comparative cost-effectiveness study. J R Coll Physicians Lond. 1993;27: Bowen J, Yaste C. Effect of stroke protocol on hospital costs of stroke patients. Neurology. 1994;44: Terént A, Lars-Åke M, Asplund K, Norrving B, Jonsson E, Wester P-O. Costs of stroke in Sweden: a national perspective. Stroke. 1994;25: Gladman J, Whynes D, Lincoln N. Cost comparison of domiciliary and hospital-based stroke rehabilitation. Age Ageing. 1994;23: Webb D, Fayad P, Wilbur C, Thomas A, Brass LM. Effects of specialized team on stroke care: the first two years of Yale Stroke Program. Stroke. 1995;26: Smurrawska L, Alexandrov A, Bladin C, Norris J. Cost of acute stroke in Toronto, Canada. Stroke. 1994;25: Hui E, Lum C, Woo J, Kay RLC. Outcomes of elderly stroke patients: day hospital versus conventional medical management. Stroke. 1995;26: Bergman L, van der Meulen JH, Limburg M, Habbema DF. Costs of medical care after first-ever stroke in the Netherlands. Stroke. 1995;26: Hass U, Persson J, Brodin H, Fréden-Karlsson I, Olsson J-E, Berg I. Assessments of rehabilitation technologies in stroke. Int J Technol Assess Health Care. 1995;11:2: Taylor TN, Davis PH, Torner JC, Holmes J, Meyer JW, Jacobson MF. Lifetime cost of stroke in United States. Stroke. 1996;27: Wentworth DA, Atkinson RP. Implementation of an acute stroke program decreases hospitalization costs and length of stay. Stroke. 1996;27: Holloway R, Witter D, Lawton K, Lipscomb J, Samsa G. Inpatient costs of specific cerebrovascular events at five academic medical centers. Neurology. 1996;46: Monane M, Kanter DS, Glynn RJ, Avorn J. Variability in length of hospitalization for stroke: the role of managed care in an elderly population. Arch Neurol. 1996;53: Chiu L, Shyu W-C, Chen T-RJ. A cost-effectiveness analysis of home care and community-based nursing homes for stroke patients and their families. J Adv Nurs. 1997;26: Evers S, Engel G, Ament A. Cost of stroke in the Netherlands from a societal perspective. Stroke. 1997;28: McNamee P, Christensen J, Soutter J, Rodgers H, Craig N, Pearson P, Bond J. Cost analysis of early supported hospital discharge for stroke. Age Ageing. 1998;27: Diringer M, Edwards D, Mattson D, Akins PT, Sheedy CW, Hsu CY, Dromerick AW. Predictors of acute hospital costs of treatment of ischemic stroke in an academic center. Stroke. 1999;30: Beech R, Rudd A, Tilling K, Wolfe CD. Economic consequences of early inpatient discharge to community-based rehabilitation for stroke in an Inner-London teaching hospital. Stroke. 1999;30: Fagerberg B, Claesson L, Gosman-Hedström G, Blomstrand C. Effect of acute stroke unit care integrated in a care continuum vs conventional treatment: a randomized 1-year study of elderly patients: the Göteborg 70 Stroke Study. Stroke. 2000;31: Strand T, Asplund K, Eriksson S, Hägg E, Lithner F, Wester PO. A non-intensive stroke unit reduces functional disability and the need for long term hospitalization. Stroke. 1985;16: Drummond MF, O Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. Oxford, UK: Oxford University Press; Pocook SJ. Clinical Trials. Chichester, UK: John Wiley & Sons; Indredavik B, Bakke F, Solberg R, Rokseth R, Lund Haaheim L, Holme I. Benefit of a stroke unit: a randomized controlled trial. Stroke. 1991; 22: Scandinavian Stroke Study Group. Multicenter trial of hemodilution in ischemic stroke: background and study protocol. Stroke. 1985;16: Mahoney FI, Barthel DW. Functional evaluation: the Barthel index. Md State Med J. 1965;14: Vardeberg K, Kolsrud M, Laberg T. The Sunnaas index of ADL. World Fed Occup Ther Bull. 1991;24: Korpelainen JT, Niilekselä E, Myllylä VV. The Sunnaas index of activities of daily living: responsiveness and concurrent validity in stroke. Scand J Occup Ther. 1997;4: Hunt S, McKenna S, McEwen J, Backet E, Williams J. A quantitative approach to perceived health status: a validation study. J Epidemiol Community Health. 1980;32: Ebrahim S, Barer D, Nouri F. Use of Nottingham Health Profile with patients after stroke. J Epidemiol Community Health. 1986;40: Wiklund I. NHP: Swedish Version of Nottingham Health Profile [in Swedish]. Göteborg, Sweden: Ofta Grafiska; The FASS 1996: The Catalogue of Pharmaceutical Specialities in Sweden Johannesson M, Fagerberg B. A health-economic comparison of diet and drug treatment in obese men with mild hypertension. J Hypertens. 1992; 10: Johannesson M, Agewall S, Hartford MTH, Fagerberg B. The costeffectiveness of cardiovascular multiple-risk-factor intervention programme in treated hypertensive men. J Intern Med. 1995;237: Bland M. An Introduction to Medical Statistics. Oxford, UK: Oxford University Press; Altman DG. Practical Statistics for Medical Research. London, UK: Chapman and Hall; Stroke Unit Trialists Collaboration. Organised inpatient (stroke unit) care for stroke (Cochrane review). In: The Cochrane Library, issue 1, Oxford, UK: Update Software. 45. Wade DT, Langton Hewer R. Functional abilities after stroke: measurement, natural history and prognosis. J Neurol Neurosurg Psychiatry. 1987;50: Caro JJ, Huybrechts KF. Stroke treatment economic model (STEM): predicting long-term costs from functional status. Stroke. 1999;30: